Research On Key Technologies Of Motorcycle Intelligent Design Based On Soft Computing

China is one of the major countries of producing motorcycle. The development of intelligent design system for motorcycle is very valuable and significant to enhance productivity, decrease production cost and promote the international competitiveness of China. At the design process of motorcycle, the development of motorcycle needs not only a large number of expertises but also perfect technique of solving problem. To introduce intelligent design technology into the product design is beneficial to enhance the level of design automation and intelligence, and to improve the product quality, reduce production costs, and shorten the development cycle of product. In this dissertation, the knowledge-based intelligent design system of motorcycle is presented, which is integrated with the artificial intelligent technology. Also, several key technologies of intelligent design are analyzed based on the development procedure, and applied into the intelligent design system of motorcycle. Thesis mainly uses the knowledge engineering to set up knowledge base system, and implements knowledge-based parametric modeling of motorcycles, which provide the finite element analysis model. Based on the structural analysis, the introduction of genetic algorithm in the soft computing theories and methods implements global optimization of motorcycle structure parameters. Neural network technology is used to acquire the hidden knowledge from the results to realize the effective reasoning and application of knowledge. And rough set theory is applied to reduce and unitize the experiential knowledge to find out the important attributes affecting the decision-making and extract the expert design knowledge from the simulation and experimental results. According to these studies, the intelligent design system of motorcycle is built up to provide an effective tool for the plan design and performance analysis of motorcycle.The major contents of the paper are the followings.1) According to the available study achievements and the design procedure of the intelligent design system of motorcycle, the general design idea of the system is analyzed, and the structure of the system is presented. The system is divided into four modules, knowledge-based systems, CAD, CAE and soft computing, in order to finish the selection of design plan, parametric modeling, the simulation analysis and performance optimization and knowledge acquisition and so on.2) In accordance with the variety and complexity of the expertise of motorcycle, the knowledge model integrated with the symbol and connection model is presented. And the study emphasis is placed on the oriented-object knowledge representation. Based on the representation method, the knowledge base and its management system, and the application mechanism integrated with several reasoning method are constructed to realize the effective application of the knowledge.3) Aim to the design plan from Knowledge Base system, in order to set up knowledge-based automated design system for motorcycles, functional structure modeling method, feature-based parametric modeling technology and object-oriented method, are utilized to develop parametric design systems based on SolidWorks for implementation of motorcycle automated design.4) Based on the finite element structural analysis, since the other methods of optimizing design shall be required to lead, easy to fall into local optimization, the genetic algorithm in soft computing theory is adopted to achieve the global optimization, which is not dependent on method of derivative. Global optimization of motorcycle structural parameters is implemented.5) In order to reduce the experimental cost and acquire effectively the hidden knowledge from the simulation and experimental results, the neural network technology is introduced into the intelligent system. After the neural network is trained repeatedly, the nonlinear regression equation of the conditions and results of experimental and analysis can be built according to the connecting weights of neural network, which reflect the inner laws of the input and output patterns. Meanwhile, the knowledge is stored in the connecting weights and biases. So, on the basis of the structure of neural network, the intelligent design system of sugarcane harvester can respond to the new design conditions quickly.6) It is difficult to explain the knowledge stored in the connecting weights and biases of neural network so that rough set theory is utilized to extract the potential knowledge and reveal the inner laws hiding behind the simulation and experimental results. Through the reductions of attributes and values, the important attributes affecting the policy decision can be found out and the expertise can be acquired. Therefore, the reasoning procedure of neural network can be explained and the designer can be provided with policy decision support.